Pith. sign in

REVIEW 3 major objections 9 minor 91 references

Loop equations uniquely pin down random matrix statistics

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

For rational beta>0, the Sine-beta and Airy-beta point processes are the unique solutions of the bulk and edge loop equation hierarchies respectively.

T0 review reviewed 2026-07-09 challenge →

load-bearing objection Loop equations uniquely characterize Sine_beta and Airy_beta for rational beta, via linearization through deformed CMS systems the 3 major comments →

arxiv 2607.07617 v1 pith:K4IKOPTB submitted 2026-07-08 math.PR math-phmath.MP

Loop Equations Characterize Random Matrix Statistics

Paul Bourgade , Jiaoyang Huang This is my paper
classification math.PR math-phmath.MP MSC 60B2060G5582B44
keywords random matrix theoryloop equationsSine_beta processAiry_beta processuniversalitybeta-ensemblesCalogero-Moser-Sutherland systemsStieltjes transform
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper proves that the universal local point processes of random matrix theory — Sine_beta in the bulk and Airy_beta at the edge — are the unique solutions of their respective loop equation hierarchies, for every rational beta > 0. Loop equations are identities satisfied by the Stieltjes transform of the eigenvalue point process, derived from integration by parts in the logarithmic gas. The authors show that any point process whose Stieltjes transform satisfies these universal microscopic identities must have the same law as Sine_beta (bulk) or Airy_beta (edge). The proof proceeds in three stages: first, the loop equations force the Stieltjes transform to concentrate around its deterministic value (i in the bulk, sqrt(w) at the edge) with sub-Gaussian tails — this is a local law derived from the equations themselves, not an external input. Second, the nonlinear loop hierarchy is linearized by passing to exponential observables (averages of products of ratios of linear factors over the random configuration), which satisfy a system of linear PDEs related to deformed Calogero-Moser-Sutherland integrable systems with two species of variables. Third, this linear system is solved completely: its solution space has dimension 2^{n+m}, indexed by sign patterns, and the physical solution is selected by matching asymptotic behavior at infinity. Once the exponential observables are uniquely determined, all correlation functions of the point process are recovered by differentiation, identifying the process. The rationality of beta enters because the linearization requires a half-plane balance condition that forces beta/2 to be a ratio of integers. Whether the characterization extends to irrational beta is stated as a conjecture. As applications, the authors derive bulk universality for Wigner matrices and for random d-regular graphs with polylogarithmic degree, and simplify the proof of edge universality for fixed-degree random d-regular graphs, by verifying approximate loop equations rather than comparing to an exactly solvable reference ensemble.

Core claim

The central discovery is that the loop equation hierarchy — a BBGKY-type system of identities for the Stieltjes transform of a log-gas, obtained by integration by parts — does not merely hold for the universal random matrix limits but uniquely characterizes them. The mechanism is a three-step chain: (1) the loop equations self-consistently imply a local law (concentration of the Stieltjes transform); (2) the nonlinear hierarchy linearizes on exponential observables that satisfy deformed Calogero-Moser-Sutherland differential equations; (3) the solution space of these linear PDEs has dimension exactly 2^{n+m}, and the physical branch is selected by asymptotic matching. This turns the loop方程从被

What carries the argument

Exponential observables (expectations of principal-value products of ratios of linear factors over the point configuration), which linearize the nonlinear loop hierarchy into a system of second-order linear PDEs related to deformed Calogero-Moser-Sutherland operators with two species of variables. The solution space is classified via Groebner basis / D-module arguments (local holonomic rank 2^{n+m}), and the physical branch is selected by a Volterra fixed-point argument along shifted rays.

Load-bearing premise

The entire uniqueness argument depends on constructing exponential observables that linearize the loop hierarchy, and this construction requires beta to be rational (specifically, a half-plane balance condition forces beta/2 to be a ratio of two integers). Whether the loop equations characterize the process for irrational beta remains open.

What would settle it

If one could exhibit a point process, distinct from Sine_beta, whose Stieltjes transform satisfies the bulk loop equation hierarchy (Assumption 1.4) for some rational beta > 0, the main theorem would be false. Equivalently, the theorem predicts that any such process must have the Sine_beta law, so a single counterexample would refute it.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Universality proofs for random matrix models reduce to verifying approximate loop equations, which often follow from local laws and integration by parts — bypassing the need for comparison with exactly solvable ensembles or Gaussian-divisible approximations.
  • The characterization provides a criterion analogous to Stein's method: just as the Gaussian distribution is characterized by the integration-by-parts identity E[Xf(X)] = E[f'(X)], the Sine_beta and Airy_beta processes are characterized by their loop equation hierarchies.
  • The connection to deformed Calogero-Moser-Sutherland systems suggests that integrable structures underlying beta-ensembles persist in the microscopic scaling limit, and the beta <-> 4/beta duality of the differential system reflects a known duality in random matrix theory.
  • The conjecture that characterization holds for irrational beta would, if true, remove the main technical limitation of the current approach.
  • Bulk universality for random d-regular graphs is extended to degrees growing as slowly as (log N)^{24}, and the method suggests a path toward the conjectured universality for fixed degree d >= 3.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 9 minor

Summary. This paper proves that, for every rational β>0, the Sine_β point process is the unique particle-generated Nevanlinna solution of the bulk loop equation hierarchy (Assumption 1.4), and the Airy_β point process is the unique solution of the edge loop equation hierarchy (Assumption 1.10). The argument proceeds in three steps: (1) the loop equations imply a local law (concentration of the Stieltjes transform around ±i in the bulk and √w at the edge, with sub-Gaussian tails); (2) exponential observables—averages of products of ratios of linear factors over the point configuration—linearize the nonlinear loop hierarchy into a deformed Calogero–Moser–Sutherland (CMS) system under a half-plane balance condition (5.2) that forces β to be rational; (3) the solution space of the deformed CMS system is classified via D-module/Gröbner basis arguments (dimension 2^{n+m}) and Volterra fixed-point construction of sectorial solutions, and the physical branch is identified by matching asymptotics. The authors apply their characterization to Wigner matrices (recovering bulk universality) and to random d-regular graphs (proving bulk universality for d≫(log N)^{24} and simplifying the edge universality argument for fixed d).

Significance. This is a substantial and potentially field-defining contribution. The idea that loop equations—analogue of Stein's method identities—do not merely hold for the limiting processes but characterize them is conceptually new and powerful. The paper ships a complete, self-contained proof across 130+ pages, with the D-module classification (Sections 7–8), the explicit β=2 solutions (Appendices D–E), and the applications to random regular graphs all carried out in full detail. The convergence criteria (Theorems 1.9, 1.12) provide a concrete, falsifiable route to universality: one verifies approximate loop equations rather than comparing to a reference ensemble. The extension to random d-regular graphs in the polylogarithmic degree regime demonstrates the method's scope beyond invariant ensembles. The rationality restriction on β is a genuine but acknowledged limitation, clearly stated as a feature of the proof method rather than the expected truth, with Conjecture 1.13 formulating the expected generalization.

major comments (3)
  1. Section 6.3, Step 3 (Eq. 6.27–6.31): The identification of the coefficient c_ε = 1 for each admissible ε ∈ E_η relies on comparing the asymptotic of F(t,s) from Proposition 4.1 (Eq. 6.27) with the distinguished branch F_{ε,η} (Eq. 6.31) along a specific block-separated configuration (6.24). The argument that all non-distinguished branches F_{ε',η} with ε' ≠ ε are exponentially suppressed (Eq. 6.34) uses the estimate |algebraic part of Φ_{ε',η}| ≤ (CΛ)^{4(n+m)^2} (Eq. 6.33). Since the exponential suppression is e^{-2(Λ²+Λ)} and the algebraic growth is polynomial in Λ, this is indeed sufficient. However, the claim that the algebraic factors between different blocks are 'bounded by CΛ^4' with at most (n+m)^2 such factors deserves more justification: the exponents α_{ij}, γ_{ab}, ρ_{ia} can be negative (e.g., α_{ij} = -2/β when ε_i ≠ ε_j), so cross-block algebraic factors could in principle衰
  2. Proposition 2.18 (Eq. 2.107): The approximate edge loop equations for random d-regular graphs with fixed degree d are stated but not proved in this paper—the proof is deferred to [52]. The claim is that these equations, combined with Theorem 1.12, simplify the edge universality proof. However, the error term in (2.107) is O(N^{-(p+1)/3 - 10c}), and it is not immediately clear from the statement alone how this error scales relative to the (1 + E[R_N^{p+1}]) factor that appears in the convergence criterion (1.6). The authors should verify that the error structure in (2.107) is compatible with the hypotheses of Theorem 1.12, or explicitly state the correspondence.
  3. Section 5.1, Proposition 5.1: The half-plane balance condition (5.2) requires n = βm/2, which with β/2 = p/q gives n = kp, m = kq. The master identity (5.14) is derived under the condition (5.12) that base-point factors cancel. The cancellation in (5.15) is verified for the bulk case. For the edge case (Section 5.2, Proposition 5.3), the Airy-regularized product (5.23) involves the formal regularized product Γ_x(z), and the cancellation argument in (5.24) is stated more briefly. Given that the edge observables involve additional exponential factors e^{t/a_j} and e^{s/a_j} (from the Airy regularization), the authors should confirm that the base-point cancellation in the edge case is fully rigorous and not merely formal, particularly regarding the convergence of the infinite product.
minor comments (9)
  1. The paper is very long (130+ pages). While the length is largely justified by the completeness of the treatment, a brief roadmap at the start of Section 7 (Solutions of the bulk deformed CMS operators) explaining the logical flow: Gröbner basis (7.1) → conjugated system (7.2) → Volterra fixed point (7.3) would help the reader.
  2. In Proposition 4.1, the quantity L in (4.4) involves a sum of logarithms. The condition L ≤ L_0 in part (ii) is used to control the Taylor expansion of E[e^Ξ]. It would help to state explicitly what L_0 depends on (p, q, k) and give a rough sense of its size.
  3. The notation β/2 = p/q ∈ Q_{>0} is introduced in Proposition 4.1 but the relationship n = kp, m = kq is only made explicit later (e.g., in Corollary 4.2 and Proposition 6.3). Stating this earlier would improve readability.
  4. In the proof of Theorem 1.5 (Section 6.2), the diagonal specialization (6.12) sets w_α = t_1^(α) = ··· = s_q^(α). The claim that the observable equals 1 after this specialization should note that this follows from β/2 = p/q, which makes the numerator and denominator powers match. This is implicit but worth stating.
  5. Section 2.4.1: The condition d ≫ (log N)^{24} is stated as sufficient but 'not optimal.' The exponent 24 appears to come from the interplay between the local law (Theorem 2.12, requiring d ≥ (log N)^4) and the switching calculus error bounds. A brief remark explaining the source of the exponent 24 (rather than, say, 4 or 8) would be helpful.
  6. Typographical: In equation (2.50), the last term on the right-hand side has a factor w/(N(πϱ_E)^2), but the subsequent absorption argument uses |w| ≤ C_K. This is correct but the bound could be stated more explicitly as |w| ≤ C_K since w ∈ K.
  7. The reference to [52] for the proof of Proposition 2.18 should clarify whether the proof in [52] covers the full approximate loop equation hierarchy (all p ≥ 1) or only specific cases.
  8. In Section 7.3, Proposition 7.4, the construction of the direction ω depends on the ordering of real parts of z_i. It would be useful to note that this construction is purely deterministic and that ω depends only on z, not on the sign pattern (ε,η).
  9. Appendix F is referenced in Remark 1.7 as proving equivalence between loop equations and the BBGKY hierarchy, but the content of Appendix F is not visible in the provided text (it appears to be truncated). The authors should verify that this appendix is complete in the final version.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for a careful reading and for identifying three points where the manuscript can be improved. All three comments are substantive and concern technical details in Sections 5–6 and the application to random regular graphs in Section 2.4. We address each below. In brief: (1) the cross-block algebraic estimate in Step 3 of Proposition 6.3 needs a more careful argument handling negative exponents, and we will add the missing details; (2) the compatibility of the error term in Proposition 2.18 with Theorem 1.12 should be stated explicitly, and we will add a remark; (3) the base-point cancellation in the edge case (Proposition 5.3) is rigorous but the argument is stated too briefly, and we will expand it. We classify this as a partial revision.

read point-by-point responses

  1. Referee: Section 6.3, Step 3 (Eq. 6.27–6.31): The identification of c_ε = 1 relies on comparing asymptotics. The claim that cross-block algebraic factors are 'bounded by CΛ^4' with at most (n+m)^2 such factors deserves more justification, since exponents α_{ij}, γ_{ab}, ρ_{ia} can be negative (e.g., α_{ij} = -2/β when ε_i ≠ ε_j), so cross-block algebraic factors could in principle grow.

    Authors: The referee is correct that the bound on cross-block algebraic factors in the non-distinguished branches requires more careful justification than what is currently written. We explain the complete argument here and will incorporate it into the revision. The key observation is that for the distinguished branch ε, within each block all t-variables share the same ε-sign and all s-variables share the same η-sign, with ε-signs and η-signs opposite. Hence α_{ij}=0, γ_{ab}=0, ρ_{ia}=0 for within-block pairs. For cross-block pairs between blocks α≠β with τ_α≠τ_β, the block separation is |w_α - w_β| ≍ Λ^4 while within-block fluctuations are O(Λ). So each cross-block ratio (t_i^{(α)} - t_j^{(β)})/(w_α - w_β) = 1 + O(Λ^{-3}), and similarly for mixed and s-s cross-block differences. For the non-distinguished branch ε'≠ε, the mismatches m_α(ε')≥1 cause some within-block pairs to have ε'_i≠ε'_j, giving exponent α_{ij}=-2/β. But these within-block differences are |t_i^{(α)} - t_j^{(α)}| ≍ Λ (since the reference points ξ_r are pairwise distinct), so each such factor contributes Λ^{-2/β}, which is polynomial in Λ. The total number of such factors is at most (n+m)^2. Crucially, the total power of (w_α - w_β) from cross-block factors cancels (as verified in the manuscript: -2p²/β - βq²/2 + 2pq = 0 using β/2=p/q), so cross-block factors contribute 1+O(Λ^{-3}). The within-block factors with negative exponents contribute at most (CΛ)^{4(n+m)^2} in total, which is polynomial. Since the exponential suppression is e^{-2(Λ²+Λ)}, this polynomial growth is indeed negligible. We will add this detailed accounting to the proof. revision: yes

  2. Referee: Proposition 2.18 (Eq. 2.107): The approximate edge loop equations for random d-regular graphs with fixed degree d are stated but not proved in this paper. The error term is O(N^{-(p+1)/3 - 10c}), and it is not immediately clear how this error scales relative to the (1 + E[R_N^{p+1}]) factor in the convergence criterion (1.6). The authors should verify compatibility or explicitly state the correspondence.

    Authors: The referee raises a valid point about the compatibility between the error structure in Proposition 2.18 and the hypotheses of Theorem 1.12. We clarify the correspondence here and will add it explicitly in the revision. In Proposition 2.18, the observable is expressed in terms of Δ(z) = m_N(z) - m_d(z), which under the edge scaling N^{1/3}A^{-2/3}Δ(E + w/(AN)^{2/3}) converges to s(w) - √w. The error term O(N^{-(p+1)/3 - 10c}) in (2.107) is an absolute error, while the convergence criterion (1.6) requires the error to be o_N(1)·(1 + E[R_N^{p+1}]). Under the edge local law for random d-regular graphs (established in [52]), R_N = O(N^c) on the event Ω_N, so E[R_N^{p+1}] is polynomially bounded. The exponent -(p+1)/3 - 10c is negative for any fixed p≥1 and sufficiently small c>0, so the absolute error is o_N(1). Moreover, on Ω_N^c (which has probability O(N^{-(1-c)})), the trivial resolvent bound gives R_N = O(N^{2/3+c}), and the contribution is absorbed using the small probability. We will add a remark after Proposition 2.18 explicitly stating this correspondence and verifying that the hypotheses of Theorem 1.12 are satisfied. revision: yes

  3. Referee: Section 5.1, Proposition 5.1 / Section 5.2, Proposition 5.3: The base-point cancellation in the edge case (5.24) is stated more briefly than the bulk case (5.15). Given that edge observables involve additional exponential factors e^{t/a_j} and e^{s/a_j} from Airy regularization, the authors should confirm that the base-point cancellation is fully rigorous, particularly regarding convergence of the infinite product.

    Authors: We agree that the base-point cancellation in the edge case is stated too briefly and that the convergence of the Airy-regularized product deserves explicit verification. The cancellation is in fact rigorous, for the following reason. The formal regularized product Γ_x(z) = e^{c_0 z} ∏_{j≥1} (z - x_j) e^{z/a_j} is not absolutely convergent by itself, but the ratios appearing in (5.16) are well-defined. Specifically, under the balance condition (5.17), the base-point factors Γ_x(ε(w_ℓ)i)^{a_ℓ} cancel in the product ∏_ℓ Γ_x(w_ℓ)^{a_ℓ}, so the observable depends only on the ratios Γ_x(w_ℓ)/Γ_x(ε(w_ℓ)i). Each such ratio can be written as exp(-∫_{ε(w_ℓ)i}^{w_ℓ} s(u) du), which is well-defined since s is holomorphic on each half-plane. The exponential factors e^{t/a_j} and e^{s/a_j} from the Airy regularization appear in the per-particle factors (t - x_j)e^{t/a_j} and (s - x_j)e^{s/a_j}, and their contribution to the ratio is exp(a_ℓ(w_ℓ - ε(w_ℓ)i)/a_j) per particle. Summing over j, the total exponential factor is exp(a_ℓ(w_ℓ - ε(w_ℓ)i)·∑_j 1/a_j), which diverges. However, this divergence is exactly canceled by the c_0 term: the sum ∑_{j≥1} 1/a_j is regularized by the Airy function zeros, and the identity c_0 = Ai'(0)/Ai(0) = -∑_{j≥1} 1/a_j (in the regularized sense) ensures cancellation. More precisely, the representation (3.11) gives s(w) = ∑_{j≥1}(1/(x_j - w) - 1/a_j) - c_0, so that ∫ s(u) du = -log Γ_x(w) + log Γ_x(ε(w)i), and the integral is absolutely convergent by the rigidity estimate (3.10). We will expand the argument in Section 5.2 to make this explicit, including a verification that the rigidity estimate (3.10) ensures the integrability of s(u) - √u along the integration path, which is the key input for the convergence. revision: yes

Circularity Check

0 steps flagged

No significant circularity found. The derivation is self-contained: loop equations are stated independently of the target processes, and all consequences flow from them.

full rationale

The paper's central claim—that Sine_β (resp. Airy_β) is the unique particle-generated Nevanlinna solution of the bulk (resp. edge) loop equation hierarchy—is derived through a chain that does not reduce to its inputs by construction. The loop equations (Assumptions 1.4, 1.10) are stated as abstract conditions on s(w), independent of Sine_β or Airy_β. From these, the paper derives: (1) the local law (Proposition 3.3, proven in §3.4 via a moment bootstrap with explicit test functions and Young's inequality), (2) exponential observables satisfying the deformed CMS system (Proposition 5.1, derived from the loop equations via the master identity (5.14)), (3) classification of the CMS solution space (Theorem 6.2, proven via Gröbner basis/D-module arguments in §7), and (4) identification of the physical branch by matching asymptotics (Proposition 6.3, using decay and boundedness estimates from Corollary 4.2, which itself derives from the concentration estimates of Proposition 3.3). The Sine_β and Airy_β processes appear only as the target of the characterization (Theorems 3.1, 3.2 show they satisfy the loop equations, proven by scaling limits of β-ensemble loop equations), not as inputs to the uniqueness argument. The correlation functions are recovered from the uniquely determined exponential observable by differentiation at a diagonal specialization (6.12), which is a standard analytic operation, not a definitional equivalence. The only self-citation is to [19] for the moment bootstrap method used in the local law proof, but the proof is carried out in full detail within the paper (§3.4), making this citation non-load-bearing. The rationality restriction on β is an acknowledged limitation of the method (the half-plane balance condition (5.2) requires β/2 ∈ ℚ), not a circularity concern. The derivation is self-contained against external benchmarks, as evidenced by the applications to Wigner matrices (§2.3) and random d-regular graphs (§2.4), where the approximate loop equations are verified independently and universality follows as a consequence. Score 1 reflects the minor, non-load-bearing self-citation to [19].

Axiom & Free-Parameter Ledger

0 free parameters · 4 axioms · 0 invented entities

The paper introduces no new physical entities or free parameters. The key mathematical objects (loop equations, deformed CMS operators, exponential observables) are derived from the loop equations, not postulated. The rationality condition on beta is an ad hoc restriction needed for the proof, not a fundamental axiom.

axioms (4)
  • domain assumption The loop equations (1.2) and (1.5) hold for the limiting point process, derived as scaling limits of finite-N beta-ensemble loop equations.
    This is the central assumption. The loop equations are derived from integration by parts for beta-ensembles and their scaling limits are taken in Section 3.3.
  • ad hoc to paper beta > 0 is rational, so beta/2 = p/q for positive integers p, q.
    This restriction enters through the half-plane balance condition (5.2) and is needed to construct the exponential observables that linearize the loop hierarchy. The authors conjecture it is not essential (Conjecture 1.13).
  • domain assumption The potential V for beta-ensembles satisfies the one-cut regular condition (Assumption 3.5).
    Needed for the approximate loop equations for beta-ensembles (Propositions 3.6, 3.7) and the derivation that Sine_beta and Airy_beta satisfy the loop equations.
  • standard math Standard results on Nevanlinna functions, Helffer-Sjostrand formula, and Groebner basis theory for D-modules.
    Used throughout for the analytic framework: Nevanlinna representation (Section 2.1), Helffer-Sjostrand (Section 3.4.2), Groebner basis (Section 7.1).

reviewed 2026-07-09 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Loop Equations Characterize Random Matrix Statistics." pith.science (2026). https://pith.science/paper/K4IKOPTB

@misc{pith2026260707617,
  author       = {Pith},
  title        = {Pith review of: Loop Equations Characterize Random Matrix Statistics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/K4IKOPTB}},
  note         = {Machine review of arXiv:2607.07617}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

We prove that the universal local point processes of random matrix theory are characterized by their loop equation hierarchies. More precisely, for every rational $\beta>0$, the $\mathrm{Sine}_{\beta}$ point process is the unique solution of the bulk loop equation hierarchy, and the $\mathrm{Airy}_{\beta}$ point process is the unique solution of the edge loop equation hierarchy. These uniqueness results provide a direct route to universality: it suffices to verify the corresponding approximate loop equations for the ensemble. In many models, these equations follow from local laws and integration by parts.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

91 extracted references · 91 canonical work pages · 3 internal anchors

  1. [1]

    Aizenman and S

    M. Aizenman and S. Warzel. On the ubiquity of the Cauchy distribution in spectral problems.Probab. Theory Related Fields, 163(1-2):61–87, 2015

    work page 2015

  2. [2]

    Akemann, J

    G. Akemann, J. Baik, and P. Di Francesco.The Oxford handbook of random matrix theory. Oxford University Press, 2011. 138

    work page 2011

  3. [3]

    Akemann and Y

    G. Akemann and Y. Fyodorov. Universal random matrix correlations of ratios of characteristic polyno- mials at the spectral edges.Nuclear Physics B, 664(3):457–476, 2003

    work page 2003

  4. [4]

    F. Atai, M. Halln¨ as, and E. Langmann. Orthogonality of super-Jack polynomials and a Hilbert space interpretation of deformed Calogero–Moser–Sutherland operators.Bulletin of the London Mathematical Society, 51(2):353–370, 2019

    work page 2019

  5. [5]

    J. Baik, P. Deift, and K. Johansson. On the distribution of the length of the longest increasing subse- quence of random permutations.J. Amer. Math. Soc., 12(4):1119–1178, 1999

    work page 1999

  6. [6]

    J. Baik, P. Deift, and E. Strahov. Products and ratios of characteristic polynomials of random Hermitian matrices.Journal of Mathematical Physics, 44(8):3657–3670, 2003

    work page 2003

  7. [7]

    Bauerschmidt, J

    R. Bauerschmidt, J. Huang, A. Knowles, and H.-T. Yau. Bulk eigenvalue statistics for random regular graphs.The Annals of Probability, 45(6A):3626 – 3663, 2017

    work page 2017

  8. [8]

    Bauerschmidt, J

    R. Bauerschmidt, J. Huang, A. Knowles, and H.-T. Yau. Edge rigidity and universality of random regular graphs of intermediate degree.Geometric and Functional Analysis, 30(3):693–769, 2020

    work page 2020

  9. [9]

    Bauerschmidt, A

    R. Bauerschmidt, A. Knowles, and H.-T. Yau. Local semicircle law for random regular graphs.Com- munications on Pure and Applied Mathematics, 70(10):1898–1960, 2017

    work page 1960

  10. [10]

    A. F. Beardon.The geometry of discrete groups. Springer Science & Business Media, 2012

    work page 2012

  11. [11]

    Bekerman, A

    F. Bekerman, A. Figalli, and A. Guionnet. Transport maps forβ-matrix models and universality. Comm. Math. Phys., 338(2):589–619, 2015

    work page 2015

  12. [12]

    Bohigas, M.-J

    O. Bohigas, M.-J. Giannoni, and C. Schmit. Characterization of chaotic quantum spectra and univer- sality of level fluctuation laws.Physical Review Letters, 52(1):1, 1984

    work page 1984

  13. [13]

    Borodin and E

    A. Borodin and E. Strahov. Averages of characteristic polynomials in random matrix theory.Commu- nications on Pure and Applied Mathematics, 59(2):161–253, 2006

    work page 2006

  14. [14]

    Borot and A

    G. Borot and A. Guionnet. Asymptotic expansion ofβmatrix models in the one-cut regime.Comm. Math. Phys., 317(2):447–483, 2013

    work page 2013

  15. [15]

    Borot, A

    G. Borot, A. Guionnet, and K. K. Kozlowski. Large-N asymptotic expansion for mean field models with Coulomb gas interaction.International Mathematics Research Notices, 2015(20):10451–10524, 2015

    work page 2015

  16. [16]

    Bourgade, L

    P. Bourgade, L. Erd˝ os, and H.-T. Yau. Edge universality of beta ensembles.Comm. Math. Phys., 332(1):261–353, 2014

    work page 2014

  17. [17]

    Bourgade, L

    P. Bourgade, L. Erd˝ os, and H.-T. Yau. Universality of generalβ-ensembles.Duke Math. J., 163(6):1127– 1190, 2014

    work page 2014

  18. [18]

    Bourgade, L

    P. Bourgade, L. Erd˝ os, H.-T. Yau, and J. Yin. Fixed energy universality for generalized Wigner matrices. Communications on Pure and Applied Mathematics, 69(10):1815–1881, 2016

    work page 2016

  19. [19]

    Bourgade, K

    P. Bourgade, K. Mody, and M. Pain. Optimal local law and central limit theorem forβ-ensembles. Comm. Math. Phys., 390(3):1017–1079, 2022

    work page 2022

  20. [20]

    Bourgade, M

    P. Bourgade, M. Sodin, and O. Zeitouni. The Paley-Wiener excess and a local law for the Sine β process. in preparation

    work page

  21. [21]

    Bourgade and H.-T

    P. Bourgade and H.-T. Yau. The eigenvector moment flow and local quantum unique ergodicity.Comm. Math. Phys., 350(1):231–278, 2017

    work page 2017

  22. [22]

    Chalykh, M

    O. Chalykh, M. Feigin, and A. Veselov. New integrable generalizations of Calogero–Moser quantum problem.Journal of Mathematical Physics, 39(2):695–703, 1998. 139

    work page 1998

  23. [23]

    Cipolloni, L

    G. Cipolloni, L. Erd˝ os, and D. Schr¨ oder. Eigenstate thermalization hypothesis for Wigner matrices. Comm. Math. Phys., 388(2):1005–1048, 2021

    work page 2021

  24. [24]

    Howe pairs, supersymmetry, and ratios of random characteristic polynomials for the unitary groups U(N)

    J. Conrey, D. Farmer, and M. Zirnbauer. Howe pairs, supersymmetry, and ratios of random characteristic polynomials for the unitary groups U (N).arXiv preprint math-ph/0511024, 2005

    work page internal anchor Pith review Pith/arXiv arXiv 2005

  25. [25]

    I. Corwin. The Kardar–Parisi–Zhang equation and universality class.Random matrices: Theory and applications, 1(01):1130001, 2012

    work page 2012

  26. [26]

    Deift.Orthogonal Polynomials and Random Matrices: A Riemann-Hilbert Approach: A Riemann- Hilbert Approach, volume 3

    P. Deift.Orthogonal Polynomials and Random Matrices: A Riemann-Hilbert Approach: A Riemann- Hilbert Approach, volume 3. American Mathematical Soc., 2000

    work page 2000

  27. [27]

    Deift and D

    P. Deift and D. Gioev.Random matrix theory: invariant ensembles and universality, volume 18. Amer- ican Mathematical Soc., 2009

    work page 2009

  28. [28]

    Dereudre, A

    D. Dereudre, A. Hardy, T. Lebl´ e, and M. Ma¨ ıda. DLR equations and rigidity for the sine-beta process. Comm. Pure Appl. Math., 74(1):172–222, 2021

    work page 2021

  29. [29]

    Desrosiers

    P. Desrosiers. Duality in random matrix ensembles for allβ.Nuclear Physics B, 817(3):224–251, 2009

    work page 2009

  30. [30]

    Desrosiers and D.-Z

    P. Desrosiers and D.-Z. Liu. Asymptotics for products of characteristic polynomials in classicalβ- ensembles.Constructive Approximation, 39(2):273–322, 2014

    work page 2014

  31. [31]

    Desrosiers and D.-Z

    P. Desrosiers and D.-Z. Liu. Selberg integrals, super-hypergeometric functions and applications to β-ensembles of random matrices.Random Matrices: Theory and Applications, 4(02):1550007, 2015. [32]NIST Digital Library of Mathematical Functions.https://dlmf.nist.gov/, Release 1.2.1 of 2024-06-

    work page 2015

  32. [32]

    F. W. J. Olver, A. B. Olde Daalhuis, D. W. Lozier, B. I. Schneider, R. F. Boisvert, C. W. Clark, B. R. Miller, B. V. Saunders, H. S. Cohl, and M. A. McClain, eds

    work page

  33. [33]

    Erbar, M

    M. Erbar, M. Huesmann, and T. Lebl´ e. The one-dimensional log-gas free energy has a unique minimizer. Communications on Pure and Applied Mathematics, 74(3):615–675, 2021

    work page 2021

  34. [34]

    Erd˝ os, A

    L. Erd˝ os, A. Knowles, H.-T. Yau, J. Yin, et al. The local semicircle law for a general class of random matrices.Electron. J. Probab, 18(59):1–58, 2013

    work page 2013

  35. [35]

    Erd˝ os, S

    L. Erd˝ os, S. P´ ech´ e, J. A. Ram´ ırez, B. Schlein, and H.-T. Yau. Bulk universality for Wigner matrices. Communications on Pure and Applied Mathematics, 63(7):895–925, 2010

    work page 2010

  36. [36]

    Erd˝ os, B

    L. Erd˝ os, B. Schlein, and H.-T. Yau. Local semicircle law and complete delocalization for Wigner random matrices.Comm. Math. Phys., 287(2):641–655, 2009

    work page 2009

  37. [37]

    Erd˝ os, B

    L. Erd˝ os, B. Schlein, and H.-T. Yau. Universality of random matrices and local relaxation flow.Invent. Math., 185(1):75–119, 2011

    work page 2011

  38. [38]

    Erd˝ os, B

    L. Erd˝ os, B. Schlein, H.-T. Yau, and J. Yin. The local relaxation flow approach to universality of the local statistics for random matrices.Ann. Inst. Henri Poincar´ e Probab. Stat., 48(1):1–46, 2012

    work page 2012

  39. [39]

    Erd˝ os and H.-T

    L. Erd˝ os and H.-T. Yau. Gap universality of generalized Wigner andβ-ensembles.J. Eur. Math. Soc. (JEMS), 17(8):1927–2036, 2015

    work page 1927

  40. [40]

    Erd˝ os and H.-T

    L. Erd˝ os and H.-T. Yau.A dynamical approach to random matrix theory, volume 28. American Math- ematical Soc., 2017

    work page 2017

  41. [41]

    Erd˝ os, H.-T

    L. Erd˝ os, H.-T. Yau, and J. Yin. Bulk universality for generalized Wigner matrices.Probability Theory and Related Fields, 154(1):341–407, 2012

    work page 2012

  42. [42]

    Erd˝ os, H.-T

    L. Erd˝ os, H.-T. Yau, and J. Yin. Rigidity of eigenvalues of generalized Wigner matrices.Adv. Math., 229(3):1435–1515, 2012. 140

    work page 2012

  43. [43]

    Eynard and N

    B. Eynard and N. Orantin. Invariants of algebraic curves and topological expansion.Commun. Number Theory Phys., 1(2):347–452, 2007

    work page 2007

  44. [44]

    Figalli and A

    A. Figalli and A. Guionnet. Universality in several-matrix models via approximate transport maps. Acta Math., 217(1):81–176, 2016

    work page 2016

  45. [45]

    P. J. Forrester. Dualities in random matrix theory.arXiv preprint arXiv:2501.07144, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  46. [46]

    Gallavotti and E

    G. Gallavotti and E. Verboven. On the classical KMS boundary condition.Il Nuovo Cimento, 28(12):274–286, 1975

    work page 1975

  47. [47]

    Genovese and S

    G. Genovese and S. Simonella. On the stationary BBGKY hierarchy for equilibrium states.J. Stat. Phys., 148(1):89–112, 2012

    work page 2012

  48. [48]

    Guionnet.Asymptotics of random matrices and related models: the uses of Dyson-Schwinger equa- tions, volume 130

    A. Guionnet.Asymptotics of random matrices and related models: the uses of Dyson-Schwinger equa- tions, volume 130. American Mathematical Soc., 2019

    work page 2019

  49. [49]

    Z. Ha. Exact dynamical correlation functions of Calogero-Sutherland model and one-dimensional frac- tional statistics.Physical Review Letters, 73(12):1574, 1994

    work page 1994

  50. [50]

    Halln¨ as

    M. Halln¨ as. New orthogonality relations for super-Jack polynomials and an associated Lassalle–Nekrasov Correspondence.Constructive Approximation, 59(1):113–142, 2024

    work page 2024

  51. [51]

    Y. He, A. Knowles, and R. Rosenthal. Isotropic self-consistent equations for mean-field random matrices. Probability Theory and Related Fields, 171(1):203–249, 2018

    work page 2018

  52. [52]

    Ramanujan Property and Edge Universality of Random Regular Graphs

    J. Huang, T. McKenzie, and H.-T. Yau. Ramanujan property and edge universality of random regular graphs.arXiv preprint arXiv:2412.20263, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  53. [53]

    Huang and H.-T

    J. Huang and H.-T. Yau. Spectrum of random d-regular graphs up to the edge.Communications on Pure and Applied Mathematics, 77(3):1635–1723, 2024

    work page 2024

  54. [54]

    Huang and H.-T

    J. Huang and H.-T. Yau. Lecture notes on edge universality for random regular graphs.arXiv preprint arXiv:2602.00975, 2026

    work page arXiv 2026

  55. [55]

    Huang and L

    J. Huang and L. Zhang. A convergence framework for Airy β line ensemble via pole evolution.Adv. Math., 498, 2026

    work page 2026

  56. [56]

    Jakobson, S

    D. Jakobson, S. D. Miller, I. Rivin, and Z. Rudnick. Eigenvalue spacings for regular graphs. InEmerging Applications of Number Theory, pages 317–327. Springer, 1999

    work page 1999

  57. [57]

    Johansson

    K. Johansson. On fluctuations of eigenvalues of random Hermitian matrices.Duke Mathematical Journal, 91(1):151 – 204, 1998

    work page 1998

  58. [58]

    Johansson

    K. Johansson. Shape fluctuations and random matrices.Comm. Math. Phys., 209(2):437–476, 2000

    work page 2000

  59. [59]

    Johansson

    K. Johansson. Universality of the Local Spacing Distribution in Certain Ensembles of Hermitian Wigner Matrices.Comm. Math. Phys., 215(3):683–705, 2001

    work page 2001

  60. [60]

    Kerov, A

    S. Kerov, A. Okounkov, and G. Olshanski. The boundary of the Young graph with Jack edge multiplic- ities.International Mathematics Research Notices, 1998(4):173–199, 1998

    work page 1998

  61. [61]

    Kontsevich

    M. Kontsevich. Intersection theory on the moduli space of curves and the matrix Airy function.Com- munications in mathematical physics, 147(1):1–23, 1992

    work page 1992

  62. [62]

    S. G. Krantz.Function theory of several complex variables, volume 340. American Mathematical Soc., 2001. 141

    work page 2001

  63. [63]

    Landon, P

    B. Landon, P. Sosoe, and H.-T. Yau. Fixed energy universality of Dyson Brownian motion.Advances in Mathematics, 346:1137–1332, 2019

    work page 2019

  64. [64]

    Lebl´ e and S

    T. Lebl´ e and S. Serfaty. Large deviation principle for empirical fields of log and Riesz gases.Inventiones mathematicae, 210(3):645–757, 2017

    work page 2017

  65. [65]

    Lesage, V

    F. Lesage, V. Pasquier, and D. Serban. Dynamical correlation functions in the Calogero-Sutherland model.Nuclear Physics B, 435(3):585–603, 1995

    work page 1995

  66. [66]

    Y. M. Makeenko and A. A. Migdal. Exact equation for the loop average in multicolor QCD.Physics Letters B, 88(1-2):135–137, 1979

    work page 1979

  67. [67]

    P. A. Martin. Sum rules in charged fluids.Rev. Modern Phys., 60(4):1075–1127, 1988

    work page 1988

  68. [68]

    M. L. Mehta.Random matrices, volume 142. Elsevier, 2004

    work page 2004

  69. [69]

    A. A. Migdal. Loop equations and 1/Nexpansion.Physics Reports, 102(4):199–290, 1983

    work page 1983

  70. [70]

    H. L. Montgomery. The pair correlation of zeros of the zeta function. InProc. Symp. Pure Math, volume 24, 1973

    work page 1973

  71. [71]

    C. B. Morrey, Jr. On the derivation of the equations of hydrodynamics from statistical mechanics. Comm. Pure Appl. Math., 8:279–326, 1955

    work page 1955

  72. [72]

    G. Nemes. Error bounds for the large-argument asymptotic expansions of the Hankel and Bessel func- tions.Acta Applicandae Mathematicae, 150:141–177, 2017

    work page 2017

  73. [73]

    A. M. Odlyzko. On the distribution of spacings between zeros of the zeta function.Mathematics of Computation, 48(177):273–308, 1987

    work page 1987

  74. [74]

    Okounkov

    A. Okounkov. On n-point correlations in the log-gas at rational temperature.arXiv preprint hep- th/9702001, 1997

    work page arXiv 1997

  75. [75]

    Olshanetsky and A

    M. Olshanetsky and A. Perelomov. Quantum integrable systems related to Lie algebras.Physics Reports, 94(6):313–404, 1983

    work page 1983

  76. [76]

    Pr¨ ahofer and H

    M. Pr¨ ahofer and H. Spohn. Scale invariance of the PNG droplet and the Airy process.Journal of statistical physics, 108(5):1071–1106, 2002

    work page 2002

  77. [77]

    Quastel and H

    J. Quastel and H. Spohn. The one-dimensional KPZ equation and its universality class.Journal of Statistical Physics, 160(4):965–984, 2015

    work page 2015

  78. [78]

    Ramirez, B

    J. Ramirez, B. Rider, and B. Vir´ ag. Beta ensembles, stochastic Airy spectrum, and a diffusion.Journal of the American Mathematical Society, 24(4):919–944, 2011

    work page 2011

  79. [79]

    Saito, B

    M. Saito, B. Sturmfels, and N. Takayama.Gr¨ obner deformations of hypergeometric differential equations. Springer Science & Business Media, 2013

    work page 2013

  80. [80]

    Serban, F

    D. Serban, F. Lesage, and V. Pasquier. Single-particle Green function in the Calogero-Sutherland model for rational couplingsβ= p/q.Nuclear Physics B, 466(3):499–512, 1996

    work page 1996

Showing first 80 references.

This paper was first reviewed by glm-5.2 on July 9, 2026.